I am trying to figure out whether it’s possible or not to get the expected upper limit with some model when the analysis is still blind. Here is my situation.
My analysis is blind so I don’t have access to some region of the observables where I expect to be sensitive to the signal, however I can construct a model for the signal + background distribution using sidebands and simulations. I am now interested in getting an estimation of the upper limit with my model and I was wondering how I can compute the distribution of the test statistics without relying on data. Is one of the calculators available in RooStats any good for this purpose? From my understanding both the Frequentist and Asymptotic calculators require a fit on data to get the values of nuisance parameters (right?) while the HybridCalculator could do the job, am I wrong?